258 
PACIFIC SCIENCE, Vol. XV, April 1961 
Movement from the trap area could account 
for a diminution in recoveries of tagged fish; 
however, evidence is strong against at least ex- 
tensive migration by this species (see section 
on migration ) . 
The marked drop in number of recoveries still 
seems too high to be accounted by any or all of 
the above reasons. One other possibility exists 
which is supported by the extraordinary high 
percentage of early returns. A tagged manini 
may show a greater predilection to re-enter a 
trap than untagged manini to enter a first time. 
This phenomenon has been conclusively demon- 
strated by Lawrence ( 1952 ) for bluegill sun- 
fish in Iowa farm ponds. If such a tendency 
exists, it might be expected to be greatest in 
the fish soon after release and gradually abate 
with time. One manini was caught, tagged, re- 
leased near the next trap, and caught again in 
this trap a few minutes later. 
The 173 tagged fish which were returned 
alive were measured alive a second time by Ne- 
moto and released. The remaining tagged ma- 
nini were measured after they had died, and 
usually not when fresh but following partial 
drying and/or freezing. Seventeen of the latter 
group were recovered within the first 2 weeks 
following tagging. These fish were all from 1 to 
7 mm. shorter than when first measured. The 
average shrinkage was 3.53 mm. The average 
standard length of these specimens measured to 
the peduncular mark is 109.8 mm. The average 
shrinkage per millimeter of length is therefore 
0.032 mm. A correction was made for each fish 
which was measured dead (unless freshly dead) 
by multiplying its recovery length by this fac- 
tor and adding the product to the length. 
A total of 70 tagged manini were measured 
alive a second time following recovery within 
2 weeks of tagging. Surprisingly, 46 of these 
were from 1 to 5 mm. shorter than when first 
measured; 10 were the same size; 14 were from 
1 to 3 mm. larger. The average "growth” was 
— 1.06 mm. A t test applied to these data to as- 
certain the probability of obtaining such a dis- 
tribution from a population in which no actual 
change in length occurred resulted in a / value 
of 5.1. With 69 degrees of freedom this gives 
a probability of less than 0.001 of getting a 
higher t value; thus the null hypothesis of no 
significant difference is rejected. Since the same 
person measured the fish both times, the reason 
for the significantly shorter length is not apt to 
be any bias in measurement. The shorter length 
is probably due to shrinkage from starvation in 
the traps. Manini which are caught in traps may 
have been held in the traps for a week or more. 
Although both measurements are made on fish 
caught in traps, the effects of confinement would 
be expected to be greater on fish spending an- 
other period in traps within 2 weeks of their 
first. 
In order to test the hypothesis that actual 
shrinkage occurs during starvation, 17 trap- 
caught adult manini (none showed evidence of 
long sojourn in traps) , 112 to 131 mm. in stand- 
ard length, were tagged, measured, placed in two 
large aquaria, and allowed to starve. Eleven 
days later the 5 fish which survived were re- 
measured. A definite shrinkage of from 1 to 5 
mm. (mean 3.4 mm.) had taken place. The 
measurements were made by Nemo to. 
The large amount of variation ( — 5 to +3 
mm.) in the series of differences between the 
measurements made on the 70 fish when tagged 
and recovered within 2 weeks was unexpected. 
When it is realized that five different sources 
contribute to this variability, it is more com- 
prehensible. There is the possible error in the 
first measurement of a living fish. The second 
measurement of this same fish constitutes an 
independent source of error. The shrinkage due 
to starvation is a third. Some manini which are 
caught have not been in a trap more than a few 
hours and will not have shortened at all; others 
held prisoner for a week or more may have 
shrunk several millimeters. Some manini initially 
caught after a very short stay in a trap, tagged 
with a minimum of injury, and recovered a full 
2 weeks later after another brief period in a 
trap might have grown about a millimeter; thus 
growth itself probably forms a fourth source. 
An important fifth source of error in the meas- 
urement of the dead tagged fish is the variation 
in shrinkage after death. All of this variability 
has necessitated a statistical approach to the 
analysis of the data, and for this purpose the 
large number of recoveries has been most wel- 
come. 
Only those manini recovered after spending 
